How would you define and calculate Weekly Active Users (WAU)?

What's really being asked

This question assesses your product thinking and data literacy. The interviewer wants to see if you can move from a vague business concept ('active users') to a concrete, defensible technical definition. They are testing your ability to define what 'value' looks like for a user, identify key actions, and foresee the technical challenges and biases in measurement. For a senior role, simply saying 'count distinct users' is not enough; you must show you understand the nuances.

The full answer

A strong answer has three parts. First, define the terms. A 'user' is a unique human, not a bot or test account. 'Active' should be tied to a core value-providing action, not a passive one. For Slack, this might be sending a message, adding a reaction, or joining a huddle, but NOT just opening the app. Second, explain the technical calculation. This usually involves a SQL query on an event log table, like SELECT COUNT(DISTINCT user_id) FROM events WHERE event_time >= NOW() - INTERVAL '7 days' AND event_name IN ('message_sent', 'reaction_added') AND is_bot = false;. Mention using a rolling 7-day window. Third, proactively discuss pitfalls. This includes filtering out bots (e.g., via a user_type flag or naming convention), internal employees (based on email domain), and automated background sync events that don't represent user intent.

The mistakes people make

A major red flag is defining 'active' as simply 'opening the app' or 'logging in.' This is a vanity metric. It overcounts users who open the app from a notification but don't engage, and it doesn't correlate with actual product value. Another mistake is forgetting to mention COUNT(DISTINCT user_id), which would lead to counting events instead of users. Failing to mention filtering bots or internal users shows a lack of experience with real-world data, which is never perfectly clean.

What usually comes next

How would you calculate WAU/MAU stickiness? How would this definition change for a different product, like a photo-sharing app or an e-commerce site? How would you handle a user who is active in multiple Slack workspaces? (This tests your understanding of user identity models). How would you set up A/B tests to validate that your chosen 'active' events actually correlate with retention?

A concrete example

For Slack, a good 'active' event is message_sent. A bad one is app_opened. A message_sent event shows a user is contributing to a conversation, the core loop of the product. An app_opened event could just be an accidental tap. If WAU based on app_opened is 1M, but WAU based on message_sent is only 300k, it tells you that 70% of your weekly 'users' aren't actually getting the core value, which is a critical insight. The definition directly impacts business strategy.